Search

This research examines the stability of crossed-field amplifiers (CFAs) and characterizes their different modes of operation: amplification, driven oscillation, and self-excited oscillation. The CFA used in this paper is the Recirculating Planar Crossed-Field Amplifier (RPCFA), which is a high power (MW) pulsed (300 ns) amplifier that operates around 3 GHz. Initially, the RPCFA is shown to be a stable amplifier with moderate gain (5.1 dB), but by either reducing the anode-cathode (AK) gap spacing or increasing the driving current, the amplifier operation transitions from amplification to oscillation. Depending on the operating conditions, these oscillations are either driven by the input RF signal or self-excited. These self-excited oscillations can have a lower synchronization phase velocity than the maximum velocity in the electron beam, implying that slower electrons within the Brillouin hub can interact with electromagnetic modes on the RF circuit. A cold tube analysis of the RPCFA shows that the Q-factor of certain modes on the RF circuit varies significantly when the AK gap geometry of the RPCFA is altered which leads to a discrete shift in operating frequency. The operation of the RPCFA close to Hull cutoff is found to share some key features of magnetically insulated transmission line oscillators (MILO) that could also explain the dramatic frequency shift. Instantaneous phase analysis by Hilbert transforms can be used, in conjunction with the frequency and output power analysis, to determine the onset of the transition from amplification to oscillation, and to characterize the oscillation.

We present a preliminary laboratory test of a setup designed to measure Hanbury Brown and Twiss-type intensity correlations from a chaotic light source using five spectral channels simultaneously. After averaging the zero-delay correlation peaks from all channels, we obtain an improvement of the signalto-noise ratio fairly consistent with theory. The goal is to demonstrate the feasibility and scalability of this technique to improve the sensitivity of stellar intensity interferometry using optical telescopes.

Studies conducted on the experimental site of Lavalette (IRSTEA Montpellier) have shown variability in the observed agricultural yield, either attributable to spatial or temporal heterogeneities in water and nitrogen supply or to gradients of soil properties. The latter is addressed by performing a multi-depthg geophysical prospection that delivers maps of apparent electrical resistivity., Comment: in French language, GEOFCAN, UMR GEOPS, 2014, Orsay (Universit{\'e} Paris-Sud 11), France

Recent advances in machine learning and hardware have produced embedded devices capable of performing real-time object detection with commendable accuracy. We consider a scenario in which embedded devices rely on an onboard object detector, but have the option to offload detection to a more powerful edge server when local accuracy is deemed too low. Resource constraints, however, limit the number of images that can be offloaded to the edge. Our goal is to identify which images to offload to maximize overall detection accuracy under those constraints. To that end, the paper introduces a reward metric designed to quantify potential accuracy improvements from offloading individual images, and proposes an efficient approach to make offloading decisions by estimating this reward based only on local detection results. The approach is computationally frugal enough to run on embedded devices, and empirical findings indicate that it outperforms existing alternatives in improving detection accuracy even when the fraction of offloaded images is small., Comment: SEC 2024

From the observation of a diffusion path $(X_t)_{t\in [0,T]}$ on a compact connected $d$-dimensional manifold $M$ without boundary, we consider the problem of estimating the stationary measure $\mu$ of the process. Wang and Zhu (2023) showed that for the Wasserstein metric $W_2$ and for $d\ge 5$, the convergence rate of $T^{-1/(d-2)}$ is attained by the occupation measure of the path $(X_t)_{t\in [0,T]}$ when $(X_t)_{t\in [0,T]}$ is a Langevin diffusion. We extend their result in several directions. First, we show that the rate of convergence holds for a large class of diffusion paths, whose generators are uniformly elliptic. Second, the regularity of the density $p$ of the stationary measure $\mu$ with respect to the volume measure of $M$ can be leveraged to obtain faster estimators: when $p$ belongs to a Sobolev space of order $\ell>0$, smoothing the occupation measure by convolution with a kernel yields an estimator whose rate of convergence is of order $T^{-(\ell+1)/(2\ell+d-2)}$. We further show that this rate is the minimax rate of estimation for this problem., Comment: 36 pages, 0 figures

This work tackles the challenge of efficiently selecting high-performance pre-trained vision backbones for specific target tasks. Although exhaustive search within a finite set of backbones can solve this problem, it becomes impractical for large datasets and backbone pools. To address this, we introduce Vision Backbone Efficient Selection (VIBES), which aims to quickly find well-suited backbones, potentially trading off optimality for efficiency. We propose several simple yet effective heuristics to address VIBES and evaluate them across four diverse computer vision datasets. Our results show that these approaches can identify backbones that outperform those selected from generic benchmarks, even within a limited search budget of one hour on a single GPU. We reckon VIBES marks a paradigm shift from benchmarks to task-specific optimization., Comment: 9 pages, 4 figures, under review at WACV 2025

Effective models to describe the dynamics of an open cavity have been extensively discussed in the literature. In many of these models the cavity leakage to the outside is treated as a loss introduced phenomenologically. In contrast to these, we focus here on characterizing the outgoing photon using a novel approach where the outside is treated as part of the system. In such a global system, in order to separately characterize the photon inside and outside cavity, we demonstrate a first-principle derivation of a coherent cavity-reservoir coupling function for cavities with mirrors consisting of a stack of dielectric layers. In particular, we show that due to the effects induced by the multilayer nature of the cavity mirror, even in the standardly defined high-finesse cavity regime, the cavity-reservoir system description might differ from the one where the structure of the mirror is neglected. Based on this, we define a generalized cavity response function and a cavity-reservoir coupling function, which account for the longitudinal geometric structure of the cavity mirror. This allows us to define an effective reflectivity for the cavity with a multilayer mirror as if it was sitting in a well-defined effective mirror plane. We estimate the error of such a definition by considering cavities of different lengths and mirror structures. Finally, we apply this model to characterize the dynamics of a single photon produced in such a cavity and propagating outside.

Active perception enables robots to dynamically gather information by adjusting their viewpoints, a crucial capability for interacting with complex, partially observable environments. In this paper, we present AP-VLM, a novel framework that combines active perception with a Vision-Language Model (VLM) to guide robotic exploration and answer semantic queries. Using a 3D virtual grid overlaid on the scene and orientation adjustments, AP-VLM allows a robotic manipulator to intelligently select optimal viewpoints and orientations to resolve challenging tasks, such as identifying objects in occluded or inclined positions. We evaluate our system on two robotic platforms: a 7-DOF Franka Panda and a 6-DOF UR5, across various scenes with differing object configurations. Our results demonstrate that AP-VLM significantly outperforms passive perception methods and baseline models, including Toward Grounded Common Sense Reasoning (TGCSR), particularly in scenarios where fixed camera views are inadequate. The adaptability of AP-VLM in real-world settings shows promise for enhancing robotic systems' understanding of complex environments, bridging the gap between high-level semantic reasoning and low-level control.

We investigate a human-like interpretable model of video understanding. Humans recognise complex activities in video by recognising critical spatio-temporal relations among explicitly recognised objects and parts, for example, an object entering the aperture of a container. To mimic this we build on a model which uses positions of objects and hands, and their motions, to recognise the activity taking place. To improve this model we focussed on three of the most confused classes (for this model) and identified that the lack of 3D information was the major problem. To address this we extended our basic model by adding 3D awareness in two ways: (1) A state-of-the-art object detection model was fine-tuned to determine the difference between "Container" and "NotContainer" in order to integrate object shape information into the existing object features. (2) A state-of-the-art depth estimation model was used to extract depth values for individual objects and calculate depth relations to expand the existing relations used our interpretable model. These 3D extensions to our basic model were evaluated on a subset of three superficially similar "Putting" actions from the Something-Something-v2 dataset. The results showed that the container detector did not improve performance, but the addition of depth relations made a significant improvement to performance., Comment: 5 pages, This manuscript has been accepted at the Human-inspired Computer Vision (HCV) ECCV 2024 Workshop. arXiv admin note: text overlap with arXiv:2107.05319

When the memory parameter of the elephant random walk is above a critical threshold, the process becomes superdiffusive and, once suitably normalised, converges to a non-Gaussian random variable. In a recent paper by the three first authors, it was shown that this limit variable has a density and that the associated moments satisfy a nonlinear recurrence relation. In this work, we exploit this recurrence to derive an asymptotic expansion of the moments and the asymptotic behaviour of the density at infinity. In particular, we show that an asymmetry in the distribution of the first step of the random walk leads to an asymmetry of the tails of the limit variable. These results follow from a new, explicit expression of the Stieltjes transformation of the moments in terms of special functions such as hypergeometric series and incomplete beta integrals. We also obtain other results about the random variable, such as unimodality and, for certain values of the memory parameter, log-concavity., Comment: 25 pages

Detecting actions in videos, particularly within cluttered scenes, poses significant challenges due to the limitations of 2D frame analysis from a camera perspective. Unlike human vision, which benefits from 3D understanding, recognizing actions in such environments can be difficult. This research introduces a novel approach integrating 3D features and depth maps alongside RGB features to enhance action recognition accuracy. Our method involves processing estimated depth maps through a separate branch from the RGB feature encoder and fusing the features to understand the scene and actions comprehensively. Using the Side4Video framework and VideoMamba, which employ CLIP and VisionMamba for spatial feature extraction, our approach outperformed our implementation of the Side4Video network on the Something-Something V2 dataset. Our code is available at: https://github.com/SadeghRahmaniB/DEAR, Comment: 5 pages, 1 figure, 1 table, accepted at Human-inspired Computer Vision, ECCV

We propose an original technique for the HF radar estimation of the main sea state parameters by exploiting the amplitude modulation of the radar signal time series. While the classical method for ocean wave measurement is based on the second-order ocean Doppler spectrum, this alternative approach uses the slow amplitude modulation of the Bragg wave in the radar signal, which is more robust to noise than the second-order echo. We apply this method to an annual set of HF radar data in the vicinity of Toulon (Mediterranean coast of France) and compare it with the classical second-order method, using a nearby buoy and the WWIII model as ground truth. The Doppler spectrum-based method is found to be more accurate in calculating the significant wave height and the peak wave frequency while the modulation approach provides coarser estimates but can achieve longer ranges and higher temporal coverage., Comment: 12 pages, 16 figures, 2 tables

This report documents the process that led to the NSF Workshop on "Sustainable Computing for Sustainability" held in April 2024 at NSF in Alexandria, VA, and reports on its findings. The workshop's primary goals were to (i) advance the development of research initiatives along the themes of both sustainable computing and computing for sustainability, while also (ii) helping develop and sustain the interdisciplinary teams those initiatives would need. The workshop's findings are in the form of recommendations grouped in three categories: General recommendations that cut across both themes of sustainable computing and computing for sustainability, and recommendations that are specific to sustainable computing and computing for sustainability, respectively.

The Alderney Race has been identified as a future site for the development of tidal energy, due to its bidirectional strong current reaching 5 m/s during spring tides. This hydrodynamics is very difficult to measure by in situ or remote sensing means. High-frequency coastal radars can provide a synoptic and near-real-time view of such a complex circulation, but the classical processing algorithms are not adapted to the extreme situation of strongly sheared currents. We propose an improved high-resolution direction-finding technique for the azimuthal processing of such radar data. It uses phased-array systems and combines the advantages of the usual beam-forming technique to eliminate many problems related to the distortion of Doppler spectra by extreme currents. The method is evaluated with a unique data set of radar measurements at two radar frequencies (13 and 24.5 MHz) and three spatial resolutions (200, 750, and 1500 m). The radar-based surface currents are analyzed in the light of a high-resolution numerical model and also compared with in situ measurements. While high azimuthal resolution can be achieved in this way, it is shown that the typical range resolutions of 750 and 1500 m are insufficient to account for the strong spatial variations of the surface current at some specific times and locations., Comment: 26 pages, 15 figures, 3 tables

Named entity recognition (NER) stands as a fundamental and pivotal task within the realm of Natural Language Processing. Particularly within the domain of Biomedical Method NER, this task presents notable challenges, stemming from the continual influx of domain-specific terminologies in scholarly literature. Current research in Biomedical Method (BioMethod) NER suffers from a scarcity of resources, primarily attributed to the intricate nature of methodological concepts, which necessitate a profound understanding for precise delineation. In this study, we propose a novel dataset for biomedical method entity recognition, employing an automated BioMethod entity recognition and information retrieval system to assist human annotation. Furthermore, we comprehensively explore a range of conventional and contemporary open-domain NER methodologies, including the utilization of cutting-edge large-scale language models (LLMs) customised to our dataset. Our empirical findings reveal that the large parameter counts of language models surprisingly inhibit the effective assimilation of entity extraction patterns pertaining to biomedical methods. Remarkably, the approach, leveraging the modestly sized ALBERT model (only 11MB), in conjunction with conditional random fields (CRF), achieves state-of-the-art (SOTA) performance.

We study the low-frequency noise (LFN), i.e. flicker noise, also referred to as 1/f noise, in 2D networks of molecularly functionalized gold nanoparticles (NMN: nanoparticle-molecule network). We examine the noise behaviors of the NMN hosting alkyl chains (octanethiol), fatty acid oleic acids (oleylamine), redox molecule switches (polyoxometalate derivatives) or photo-isomerizable molecules (azobenzene derivatives) and we compare their 1/f noise behaviors. These noise metrics are used to evaluate which molecules are the best candidates to build in-materio reservoir computing molecular devices based on NMNs., Comment: Full paper and supporting information, revised

We consider a classical stochastic control problem in which a diffusion process is controlled by a withdrawal process up to a termination time. The objective is to maximize the expected discounted value of the withdrawals until the first-passage time below level zero. In this work, we are considering absolutely continuous control strategies in a general diffusion model. Our main contribution is a solution to the control problem under study, which is achieved by using a probabilistic guess-and-verify approach. We prove that the optimal strategy belongs to the family of bang-bang strategies, i.e. strategies in which, above an optimal barrier level, we withdraw at the highest-allowed rate, while no withdrawals are made below this barrier. Some nontrivial examples are studied numerically.

The Kramers escape problem is a paradigmatic model for the kinetics of rare events, which are usually characterized by Arrhenius law. So far, analytical approaches have failed to capture the kinetics of rare events in the important case of non-Markovian processes with long-term memory, as occurs in the context of reactions involving proteins, long polymers, or strongly viscoelastic fluids. Here, based on a minimal model of non-Markovian Gaussian process with long-term memory, we determine quantitatively the mean FPT to a rare configuration and provide its asymptotics in the limit of a large energy barrier $E$. Our analysis unveils a correction to Arrhenius law, induced by long-term memory, which we determine analytically. This correction, which we show can be quantitatively significant, takes the form of a second effective energy barrier $E'

This paper demonstrates a self-supervised approach for learning semantic video representations. Recent vision studies show that a masking strategy for vision and natural language supervision has contributed to developing transferable visual pretraining. Our goal is to achieve a more semantic video representation by leveraging the text related to the video content during the pretraining in a fully self-supervised manner. To this end, we present FILS, a novel self-supervised video Feature prediction In semantic Language Space (FILS). The vision model can capture valuable structured information by correctly predicting masked feature semantics in language space. It is learned using a patch-wise video-text contrastive strategy, in which the text representations act as prototypes for transforming vision features into a language space, which are then used as targets for semantically meaningful feature prediction using our masked encoder-decoder structure. FILS demonstrates remarkable transferability on downstream action recognition tasks, achieving state-of-the-art on challenging egocentric datasets, like Epic-Kitchens, Something-SomethingV2, Charades-Ego, and EGTEA, using ViT-Base. Our efficient method requires less computation and smaller batches compared to previous works.

Lift force of viscous origin is widespread across disciplines, from mechanics to biology. Here, we present the first direct measurement of the lift force acting on a particle moving in a viscous fluid along the liquid interface that separates two liquids. The force arises from the coupling between the viscous flow induced by the particle motion and the capillary deformation of the interface. The measurements show that the lift force increases as the distance between the sphere and the interface decreases, reaching saturation at small distances. The experimental results are in good agreement with the model and numerical calculation developed within the framework of the soft lubrication theory.

With the increasing use of neural networks in critical systems, runtime monitoring becomes essential to reject unsafe predictions during inference. Various techniques have emerged to establish rejection scores that maximize the separability between the distributions of safe and unsafe predictions. The efficacy of these approaches is mostly evaluated using threshold-agnostic metrics, such as the area under the receiver operating characteristic curve. However, in real-world applications, an effective monitor also requires identifying a good threshold to transform these scores into meaningful binary decisions. Despite the pivotal importance of threshold optimization, this problem has received little attention. A few studies touch upon this question, but they typically assume that the runtime data distribution mirrors the training distribution, which is a strong assumption as monitors are supposed to safeguard a system against potentially unforeseen threats. In this work, we present rigorous experiments on various image datasets to investigate: 1. The effectiveness of monitors in handling unforeseen threats, which are not available during threshold adjustments. 2. Whether integrating generic threats into the threshold optimization scheme can enhance the robustness of monitors., Comment: 13 pages, 5 figures, 6 tables. To appear in the proceedings of the 40th Conference on Uncertainty in Artificial Intelligence (UAI 2024)

We propose a simple method, based on an original change of variables, for the fast and accurate calculation of the second-order ocean Doppler spectrum that describes the sea echo of High-Frequency radars. A byproduct of the technique is the derivation of an improved weighting function which can be used for the inversion of the main sea state parameters. For this we revisit Barrick's method for the estimation of the significant wave height and the mean period from the ocean Doppler spectrum. On the basis of numerical simulations we show that a better estimation of these parameters can be reached but necessitates a preliminary bias correction that depends only on the radar frequency. A second consequence of our formulation is the derivation of a simple yet analytical nonlinear approximation of the second-order ocean Doppler spectrum when the Doppler frequency is larger than the Bragg frequency. This opens new perspective for the inversion of directional wave spectra from High-Frequency radar measurements., Comment: 8 pages, 8 figures, 1 table

This paper considers networks where user traffic is regulated through deterministic traffic profiles, e.g., token buckets, and requires hard delay bounds. The network's goal is to minimize the resources it needs to meet those bounds. The paper explores how reprofiling, i.e., proactively modifying how user traffic enters the network, can be of benefit. Reprofiling produces ``smoother'' flows but introduces an up-front access delay that forces tighter network delays. The paper explores this trade-off and demonstrates that, unlike what holds in the single-hop case, reprofiling can be of benefit} even when ``optimal'' schedulers are available at each hop.

Unsupervised on-the-fly back-translation, in conjunction with multilingual pretraining, is the dominant method for unsupervised neural machine translation. Theoretically, however, the method should not work in general. We therefore conduct controlled experiments with artificial languages to determine what properties of languages make back-translation an effective training method, covering lexical, syntactic, and semantic properties. We find, contrary to popular belief, that (i) parallel word frequency distributions, (ii) partially shared vocabulary, and (iii) similar syntactic structure across languages are not sufficient to explain the success of back-translation. We show however that even crude semantic signal (similar lexical fields across languages) does improve alignment of two languages through back-translation. We conjecture that rich semantic dependencies, parallel across languages, are at the root of the success of unsupervised methods based on back-translation. Overall, the success of unsupervised machine translation was far from being analytically guaranteed. Instead, it is another proof that languages of the world share deep similarities, and we hope to show how to identify which of these similarities can serve the development of unsupervised, cross-linguistic tools.

We consider the motion of a harmonically trapped overdamped particle, which is submitted to a self-phoretic force, that is proportional to the gradient of a diffusive field for which the particle itself is the source. In agreement with existing results for free particles or particles in a bounded domain, we find that the system exhibits a transition between an immobile phase, where the particle stays at the center of the trap, and an oscillatory state. We perform an exact analysis giving access to the bifurcation threshold, as well as the frequency of oscillations and their amplitude near the threshold. Our analysis also characterizes the shape of two-dimensional oscillations, that take place along a circle or a straight line. Our results are confirmed by numerical simulations., Comment: 10 pages 8 figures

Quantum optimal control is a set of methods for designing time-varying electromagnetic fields to perform operations in quantum technologies. This tutorial paper introduces the basic elements of this theory based on the Pontryagin maximum principle, in a physicist-friendly way. An analogy with classical Lagrangian and Hamiltonian mechanics is proposed to present the main results used in this field. Emphasis is placed on the different numerical algorithms to solve a quantum optimal control problem. Several examples ranging from the control of two-level quantum systems to that of Bose-Einstein Condensates (BEC) in a one-dimensional optical lattice are studied in detail, using both analytical and numerical methods. Codes based on shooting method and gradient-based algorithms are provided. The connection between optimal processes and the quantum speed limit is also discussed in two-level quantum systems. In the case of BEC, the experimental implementation of optimal control protocols is described, both for two-level and many-level cases, with the current constraints and limitations of such platforms. This presentation is illustrated by the corresponding experimental results., Comment: 51 pages, 15 figures, 272 references

Splitting probabilities quantify the likelihood of a given outcome out of competitive events for general random processes. This key observable of random walk theory, historically introduced as the Gambler's ruin problem for a player in a casino, has a broad range of applications beyond mathematical finance in evolution genetics, physics and chemistry, such as allele fixation, polymer translocation, protein folding and more generally competitive reactions. The statistics of competitive events is well understood for memoryless (Markovian) processes. However, in complex systems such as polymer fluids, the motion of a particle should typically be described as a process with memory. Appart from scaling theories and perturbative approaches in one-dimension, the outcome of competitive events is much less characterized analytically for processes with memory. Here, we introduce an analytical approach that provides the splitting probabilities for general $d$-dimensional non-Markovian Gaussian processes. This analysis shows that splitting probabilities are critically controlled by the out of equilibrium statistics of reactive trajectories, observed after the first passage. This hallmark of non-Markovian dynamics and its quantitative impact on splitting probabilities are directly evidenced in a prototypical experimental reaction scheme in viscoelastic fluids. Altogether, these results reveal both experimentally and theoretically the importance of memory effects on competitive reactions.

Domain decay is at the heart of the so-called evaporation-condensation Ostwald-ripening regime of phase ordering kinetics, where the growth of large domains occurs at the expense of smaller ones, which are expected to `evaporate'. We experimentally investigate such decay dynamics at the level of a single spherical domain picked from one phase in coexistence and brought into the other phase by an opto-mechanical approach, in a near-critical phase-separated binary liquid mixture. We observe that the decay dynamics is generally not compatible with the theoretically expected surface-tension decay laws for conserved order parameters. Using a mean-field description, we quantitatively explain this apparent disagreement by the gradient of solute concentrations induced by gravity close to a critical point. Finally, we determine the conditions for which buoyancy becomes negligible compared to capillarity and perform dedicated experiments that retrieve the predicted surface-tension induced decay exponent. The surface-tension driven decay dynamics of conserved order parameter systems in the presence and the absence of gravity, is thus established at the level of a single domain., Comment: 4 pages (main text) + 8 pages (SM); Accepted in PRL

Existing methods for evaluating large language models face challenges such as data contamination, sensitivity to prompts, and the high cost of benchmark creation. To address this, we propose a lossless data compression based evaluation approach that tests how models' predictive abilities generalize after their training cutoff. Specifically, we collect comprehensive test data spanning 83 months from 2017 to 2023 and split the data into training and testing periods according to models' training data cutoff. We measure: 1) the compression performance on the testing period as a measure of generalization on unseen data; and 2) the performance gap between the training and testing period as a measure of robustness. Our experiments test 14 representative large language models with various sizes on sources including Wikipedia, news articles, code, arXiv papers, and multi-modal data. We find that the compression rate of many models reduces significantly after their cutoff date, but models such as Mistral and Llama-2 demonstrate a good balance between performance and robustness. Results also suggest that models struggle to generalize on news and code data, but work especially well on arXiv papers. We also find the context size and tokenization implementation have a big impact of on the overall compression performance.

In this chapter, we consider the problem of a non-Markovian random walker (displaying memory effects) searching for a target. We review an approach that links the first passage statistics to the properties of trajectories followed by the random walker in the future of the first passage time. This approach holds in one and higher spatial dimensions, when the dynamics in the vicinity of the target is Gaussian, and it is applied to three paradigmatic target search problems: the search for a target in confinement, the search for a rarely reached configuration (rare event kinetics), or the search for a target in infinite space, for processes featuring stationary increments or transient aging. The theory gives access to the mean first passage time (when it exists) or to the behavior of the survival probability at long times, and agrees with the available exact results obtained perturbatively for examples of weakly non-Markovian processes. This general approach reveals that the characterization of the non-equilibrium state of the system at the instant of first passage is key to derive first-passage kinetics, and provides a new methodology, via the analysis of trajectories after the first-passage, to make it quantitative., Comment: 22 pages, invited chapter for the book "The Target Problem" (Eds. D. S. Grebenkov, R. Metzler, G. Oshanin)

Metaphors are considered to pose challenges for a wide spectrum of NLP tasks. This gives rise to the area of computational metaphor processing. However, it remains unclear what types of metaphors challenge current state-of-the-art models. In this paper, we test various NLP models on the VUA metaphor dataset and quantify to what extent metaphors affect models' performance on various downstream tasks. Analysis reveals that VUA includes a large number of metaphors that pose little difficulty to downstream tasks. We would like to shift the attention of researchers away from these metaphors to instead focus on challenging metaphors. To identify hard metaphors, we propose an automatic pipeline that identifies metaphors that challenge a particular model. Our analysis demonstrates that our detected hard metaphors contrast significantly with VUA and reduce the accuracy of machine translation by 16\%, QA performance by 4\%, NLI by 7\%, and metaphor identification recall by over 14\% for various popular NLP systems.

In this note, we study a family of polynomials that appear naturally when analysing the characteristic functions of the one-dimensional elephant random walk. These polynomials depend on a memory parameter $p$ attached to the model. For certain values of $p$, these polynomials specialise to classical polynomials, such as the Chebychev polynomials in the simplest case, or generating polynomials of various combinatorial triangular arrays (e.g.\ Eulerian numbers). Although these polynomials are generically non-orthogonal (except for $p=\frac{1}{2}$ and $p=1$), they have interlacing roots. Finally, we relate some algebraic properties of these polynomials to the probabilistic behaviour of the elephant random walk. Our methods are reminiscent of classical orthogonal polynomial theory and are elementary., Comment: 12 pages, 2 figures, 1 table